Download Full Text (1.7 MB)
Mac Johnson, CIS494: Research in Computer Information Systems
Faculty Mentor(s): Professor Sarbani Banerjee, Computer Information Systems
Person Re-Identification is the use of multiple videos or camera angles to track individuals over time. Someone walking down the street might be in several camera angles at once, but the task is to combine the videos or snapshots to show it is the same person. One problem with this idea is that pictures can have many different variations that affect the accuracy. The plan is to use multiple training libraries to find if this improves identification. Person Re-Identification is a rising topic in the Cybersecurity field and holds much potential for progress. PyCharm will be used as the IDE for this program because all the learning processes are stored in libraries in Python code. There are multiple options regarding which learning dataset to use, among them CUHK01, iLIDS-VID, and RPIField. The libraries will train the program with artificial intelligence to track a person and store images of them for future queries. Since most of the time libraries are used individually, there should be an increase in accuracy by combining them. The expectation is to see a clear advantage of training methods when used together and also identify the single most efficient library. The goal is to find a different combination of these training techniques to allow the program's artificial intelligence to be more accurate and adaptable. The results of this research project will show the differences and benefits of using multiple image libraries.
Johnson, Mac, "Person Re-Identification: Tracking the World" (2021). Computer Information Systems and Engineering Technology. 4.